Leveraging Government Data Using Unsupervised and Supervised Machine Learning for Firms’ Investment Policy-Making in Economic Crises

Abstract

In market-based economies economic crises of different geographical scopes and durations are often appearing, and are resulting in economic recessions, which have quite negative consequences for the economy and the society. Governments respond by undertaking large-scale economic stimulation programs, spending vast amounts of financial resources (with orders of magnitude between 3–6% of GDP), in order to mitigate these negative consequences. It is of critical importance to make effective use of these huge financial resources, in order to have high positive impact on the economy and the society in these tough crisis periods. This necessitates careful and rational design and implementation of these large and costly economic stimulation programs. Since one of the most important consequences of economic crises is the decrease of firms’ investments, the above economic stimulation programs include investment support actions, which aim to mitigate these crisis-induced firms’ investment decreases, and include a wide range of interventions for this reason, such as investment incentives, subsidies, low-interest loans as well as relevant tax rebates. In this paper is presented an integrated methodology for leveraging government data from economic crisis periods, using on one hand Unsupervised Machine Learning techniques, and on the other hand Supervised Machine Learning ones, in order to provide support for the rational design and implementation of firms’ investment support actions in economic crises. A first application of the proposed methodology is presented, based on existing data from the Greek Ministry of Finance and the Statistical Authority concerning 363 firms for the economic crisis period 2009–2014, which gave interesting and encouraging results.